Ironclad Inflation?

Winter break is almost upon us, and all that stands between me and the end of MATH1014 is tomorrow's Examiners' Meeting.  In keeping with my impression that the teaching procedures are more tightly regulated here, all the department instructors meet before final grades are assigned.  We've already been asked to submit our marks to the department, but we're not allowed to make even final exam scores available to the students until the College releases the grades.  At the meeting, we'll have access to the statistics from previous year's courses: pass/fail rates, average scores, number of High Distinctions, etc.  While I think that institutional memory is a valuable resource, I'm less fond of the fact that this information is more than just an optional guide.  In fact, we're not only expected to make our averages match the previous years, but also to match the College's predictions of what our students should score.  This isn't a last-minute surprise; after our students had lower than average mid-semester exam scores, we wrote a final exam designed to raise their scores so that we wouldn't have to shift them too much at the end.  I'm not entirely comfortable with the process --and to be fair, I have yet to see it in action-- but I wonder if the difference from what I'm used to is more in style than substance.

At Stanford grades were inevitable "curved", which really meant "raised".  It feels different to make such adjustments in the privacy and comfort of one's own Excel gradesheet, but for the moment I'll set aside the University's firm hand in the process.  I think that the bigger difference comes from the fact that grades here are assigned as numbers on a scale of 1 to 100, with descriptors like "Pass" and "Distinction" added afterwards.  At Stanford, I could (try to) design an exam with an average score of 60%, leaving room for an exceptional student to get 90% or the weak student to pass with 35%.  I think there's pedagogical value to an exam which delivers the message, "If you can do two-thirds of these problems, you're doing fine, but I want you to know that there's lots of room to do better."  (A topic for another post is whether this message gets through or whether such an exam is just demoralising.)  I had more flexibility in writing my exams because I assigned letter grades on my own terms.

A drawback of this approach is that it lacks transparency; students understandably found it frustrating not to know exactly where they stood throughout the semester.  This invariably worked to their advantage (see curve vs. raise above), but I think an ideal grading scheme leaves no one guessing.  Nevertheless, I don't think the ANU system is an improvement; it offers the precision of a numerical grade, but the retroactive rescaling renders this number as arbitrary as a letter.  In fact, we can't release final exam scores precisely to prevent students from being able to calculate their grades.  I'm curious to see how the meeting runs tomorrow; I suspect that most semesters, it will be a bureaucratic hassle, but at least this once it has the appeal of novelty.